Number of the records: 1
On the Over-Fitting Problem of Complex Feature Selection Methods
- 1.
SYSNO ASEP 0337806 Document Type C - Proceedings Paper (int. conf.) R&D Document Type Conference Paper Title On the Over-Fitting Problem of Complex Feature Selection Methods Title O problému přetrénování komplexních metod výběru příznaků Author(s) Somol, Petr (UTIA-B) RID
Novovičová, Jana (UTIA-B)
Pudil, Pavel (UTIA-B) RIDSource Title Proc. 5th International Computer Engineering Conference - A better Information Society Through the e@, Machine Intelligence and Web Applications. - Káhira : Cairo University, 2009 Pages s. 12-17 Number of pages 6 s. Publication form CD-ROM - CD-ROM Action 5th International Computer Engineering Conference - A better Information Society Through the e@ Event date 27.12.2009-28.12.2009 VEvent location Káhira Country EG - Egypt Event type WRD Language eng - English Country EG - Egypt Keywords feature selection ; overfitting ; overselection Subject RIV BD - Theory of Information R&D Projects 1M0572 GA MŠMT - Ministry of Education, Youth and Sports (MEYS) GA102/07/1594 GA ČR - Czech Science Foundation (CSF) GA102/08/0593 GA ČR - Czech Science Foundation (CSF) CEZ AV0Z10750506 - UTIA-B (2005-2011) Annotation One of the hot topics discussed recently in relation to machine learning is the question of actual performance of modern feature selection methods. Feature selection has been a highly active area of research in recent years due to its potential to improve both the performance and economy of automatic decision systems in various applicational fields, including medicine, image analysis, remote sensing, economics etc. The number of available methods and methodologies has grown rapidly throughout recent years while promising important improvements. Yet recently many authors put this development in question, claiming that simpler older tools are actually better than complex modern ones – which, despite promises, are claimed to actually fail in real-world applications. We investigate this question, show several illustrative examples and draw several conclusions and recommendations regarding feature selection methods’ expectable performance. Workplace Institute of Information Theory and Automation Contact Markéta Votavová, votavova@utia.cas.cz, Tel.: 266 052 201. Year of Publishing 2010
Number of the records: 1